The moral fixed points: new directions for moral nonnaturalism

Abstract

Our project in this essay is to showcase nonnaturalistic moral realism’s resources for responding to metaphysical and epistemological objections by taking the view in some new directions. The central thesis we will argue for is that there is a battery of substantive moral propositions that are also nonnaturalistic conceptual truths. We call these propositions the moral fixed points. We will argue that they must find a place in any system of moral norms that applies to beings like us, in worlds similar to our own. By committing themselves to true propositions of these sorts, nonnaturalists can fashion a view that is highly attractive in its own right, and resistant to the most prominent objections that have been pressed against it.

At some point in the mid-20th century, philosophers threw nonnaturalist moral realism on the scrap heap of philosophical theories, allowing it to gather rust and be forgotten. While the view did gather rust, it was not forgotten. To the surprise of many, this form of moral realism (henceforth, just nonnaturalism) was retrieved, repaired, and reintroduced to the mainstream metaethical discussion as a view worthy of serious consideration.¹

This reversal of fortune has not met with unanimous enthusiasm. Critics have urged that even in its most sophisticated forms, nonnaturalism remains vulnerable to some long-standing and deep-seated concerns. Some of these worries are epistemological: critics claim that our moral beliefs have not only been shaped by distorting cultural, historical, and evolutionary forces, but are also subject to deep and intractable disagreement. Given these facts, vindicating moral knowledge is a formidable undertaking, one on which nonnaturalists have made rather little headway. Other concerns about nonnaturalism target its ontological commitments: when compared to various versions of naturalism, nonnaturalism looks all too comfortable with metaphysical mysteries such as putatively brute necessary connections between moral and non-moral facts.² On this score, moral naturalism appears to do considerably better.

Our project in this essay is to showcase nonnaturalism’s resources for responding to concerns such as these by taking the view in some new directions. The central thesis we will argue for is that there is a battery of substantive moral propositions, such as 〈that it is wrong to torture others just because they have inconvenienced you〉 and 〈that it is wrong to rape a child solely to indulge one’s lust〉, that are also nonnaturalistic conceptual truths.³ These propositions, which we hold must find a place in any system of moral norms that applies to beings like us, in worlds similar to our own, we call the moral fixed points. By committing themselves to true propositions of these sorts, nonnaturalists can fashion a view that is both highly attractive and able to deflect the most prominent objections that have been leveled against it.

It is worth highlighting at the outset some of our position’s distinctive features. In the first place, our view is not intended to be “metaphysically light” but ontologically committed, implying the existence of an array of robust nonnatural truths. In this regard, our view is very different from the versions of nonnaturalism defended by “relaxed realists” such as Ronald Dworkin, Matthew Kramer, Derek Parfit, and T. M. Scanlon, which maintain that there are moral truths but that these truths have no “positive ontological implications.”⁴ Second, our preferred version of moral realism takes on semantic and conceptual commitments that go beyond even those endorsed by fellow nonnaturalists, such as David Enoch and William FitzPatrick, who share our rejection of relaxed realism.⁵ With Enoch and FitzPatrick, we endorse the thesis that there are nonnatural moral truths, but also maintain that these truths include the moral fixed points. As we see it, these truths not only constitute any reasonably comprehensive moral system for beings such as us in a world such as ours, but also fix the boundaries of moral thought: one could not engage in competent moral thinking while rejecting them.⁶ Finally, in what follows, we offer a specific account of the nature of the moral fixed points, which involves accepting a particular position regarding the nature of moral concepts and their contribution to thought. An implication of this understanding of the nature of moral concepts is that the moral fixed points are a species of conceptual truth.

Ontological commitments of these sorts, we realize, come at a cost. While we believe that the commitments of our view do not introduce new and objectionable mysteries, we nevertheless have some explaining to do. We will do this in standard ways, viz., by pointing to all of the good work done by our ontological additions. We hope to show not only that our preferred version of nonnaturalism can provide, in a pleasingly unified way, satisfying answers to challenges such as those cited earlier, which call into question the view’s epistemological and metaphysical credentials, but also that the gap between nonnaturalism and naturalism may not be as wide as many have believed.

1 Nonnaturalism characterized

We begin with a characterization of nonnaturalism as we understand it. In our view, nonnaturalism is best formulated as a thesis about moral truth. Specifically, we understand nonnaturalists to embrace:

The Core Claim: there are nonnatural moral truths,

where a nonnatural moral truth is a true moral proposition that is not identical with or made true exclusively by some natural fact. In Sects. 3 and 4, we unpack this thesis, embellishing it in various ways. For now, we note that this thesis endeavors to be neutral in two important respects.

First, The Core Claim does not incorporate any particular account of what the natural/nonnatural distinction comes to. There are various proposals in the literature, all of which face challenges. Some construe the distinction methodologically, arguing that moral inquiry is rightly undertaken in a manner that is fundamentally different from that employed in the natural sciences. Others understand the distinction metaphysically, arguing that moral properties and facts are sui generis. Still others locate the fundamental difference epistemologically, arguing that nonnaturalism is defined by a commitment to the a priority of fundamental moral norms.⁷

We prefer to focus on a metaphysical construal, and do what so many have done before us. It’s not quite punting, but neither is it anything remotely like the provision of necessary and sufficient conditions for drawing the distinction between the natural and nonnatural. The strategy is to assemble a pair of lists. The first contains only those terms that would refer to entities that most parties to these debates would agree are natural, say, because these entities are empirically knowable, or play an explanatory role in the usual sciences (or are wholly constructible from or reducible to entities that play such an explanatory role). According to this strategy, the first list would include terms that putatively refer to (i) objects such as protons and light waves, (ii) properties such as being green and being in pain, and (iii) concepts such as ‘negative charge’ and ‘being angry.’ Terms that do not make it on the first list are included on a second list, since they are live candidates for being expressions that refer to nonnatural entities. This list might include terms that putatively refer to (i) objects such as God or practical reasons, (ii) properties such as being morally wrong and being unreasonable, and (iii) concepts such as ‘being intrinsically valuable’ and ‘being sublime.’

This approach does not imply that terms on either list refer. Nor does it imply that if these terms do refer, the terms on the first list refer to natural entities, while those on the second list refer to nonnatural ones. For all we say here, Berkeleyan idealism might be true, in which case many of the terms on the first list would either not refer or would refer to nonnatural entities. Alternatively, reductive physicalism might be true, in which case many of the terms on the second list would fail to refer or would refer to natural entities. As we are envisioning their use, these lists simply play the heuristic role of designating good candidates for being natural or non-natural entities (if any there be). Officially, then, our characterization of the natural/nonnatural distinction is compatible with the possibility that the class of nonnatural entities is empty.⁸

Not all possibilities are, however, created equal. The possibility that moral properties reduce to natural ones is, in our view, genuinely open. Whether such a reduction can be achieved will turn on resolving some difficult issues regarding the correct conditions for the identity and diversity of properties. By contrast, we think it highly unlikely that moral concepts reduce to natural ones, and our view here is reinforced by the absence of any considerable disagreement on the matter.⁹ This apparent immunity to naturalistic reduction is explained (in part) by the fact that, under nearly any construal, concepts are entities with fine-grained individuation conditions. It follows that even if moral and natural concepts were necessarily co-referential—both, say, referring to natural properties, as naturalists believe—they needn’t be identical. To secure concept identity, more is needed, although we won’t venture any thoughts on what these extra conditions are or how to identify them.¹⁰ At any rate, in what follows, we’ll assume that we have excellent reason to believe that there are there moral concepts, and that some are nonnatural.

Here is a second way in which our position is relevantly neutral: while the Core Claim commits itself to the existence of nonnatural truths, it does not take a stand on whether there are any nonnatural properties or facts (assuming for the moment that true propositions are not facts). We distinguish, then:

Minimal nonnaturalism: there are nonnatural moral truths, but there are no nonnatural moral properties or facts. All moral properties and facts are natural.

from:

Robust nonnaturalism: there are both nonnatural moral truths and nonnatural moral properties and facts.

As will become evident later, we favor a view that is in the spirit (if perhaps not the letter) of Robust nonnaturalism. Still, we want to remain officially neutral on which of these two views is true. We can do this because both Minimal and Robust nonnaturalism are compatible with the primary thesis that we wish to defend, namely, that there is a range of moral propositions, the moral fixed points, which have these two distinguishing features: first, these propositions are constituted by nonnatural moral concepts, and, second, these propositions are not identical with or made true exclusively by natural facts. Rather, they are true in virtue of the nature of the nonnatural moral concepts that constitute them.

In the next two sections, we develop this thesis, sketching an account of the moral fixed points and then presenting an account of concepts according to which the moral fixed points are a species of conceptual truth.

2 Moral fixed points

There is a long history of attempts to fix the domain of moral propositions by appeal to highly formal characteristics which they are said to exhibit. For example, R. M. Hare argued that the content of a commitment is moral just so long as it is universalizable and is of overriding importance.¹¹ Contemporary “analytical functionalists” such as Frank Jackson and Philip Pettit appeal to other distinguishing marks of the moral, such as the motivational power of moral judgments and the fact that moral features supervene on descriptive characteristics.¹² Richard Joyce endorses the yet different idea that nothing could count as a moral system if it failed to include the concept of a categorical reason for action, which is assumed to have genuine application.¹³ For reasons that will emerge later, we are also interested in identifying what counts as a moral system as opposed to some other normative system. We prefer, however, to do so by taking a different route. At a first pass, we claim that something counts as a moral system only if it incorporates a sufficiently wide range of the moral fixed points.

To flesh out this idea, we begin by introducing some terminology. Let’s say that a minimally eccentric moral system is a reasonably comprehensive and consistent body of moral propositions that apply to beings like us in a world such as ours. By a reasonably comprehensive body of moral propositions, we mean an array of propositions that concern nearly all situations that agents might find themselves in and imply, often in conjunction with empirical propositions, a range of moral assessments and recommendations for those situations. A moral system, we believe, may allow for pockets of indeterminacy. But it will have the resources to provide determinate moral verdicts in many cases, and its principles can serve as deliberative guidelines for agents seeking to comport their behavior, responses, and judgments with the demands of morality.

Beings like us are mortal, embodied, not subject to daily bouts of complete amnesia,¹⁴ susceptible to physical, emotional and other psychological pleasures and pains, capable of introspection, of friendship, of self-esteem, possessed of some degree of empathy and sympathy, and able to reason deductively, inductively and abductively in at least minimal ways.

A world such as ours is one that can sustain life, one in which people do not pop in and out of existence willy-nilly, one in which the future is not known to be in all respects perfectly identical to the past, one in which material objects are observable—in short, one that is regulated by laws of nature and broad empirical statistical generalizations that are at least close to the ones that obtain here on earth.

Our proposal is that, necessarily, any minimally eccentric moral system will include an array of those propositions that we have called the moral fixed points. Here is a list of the types of propositions we have in mind:

• It is pro tanto wrong to engage in the recreational slaughter of a fellow person.

• It is pro tanto wrong to break a promise on which another is relying simply for convenience’s sake.

• It is pro tanto wrong to humiliate others simply for pleasure.

• It is pro tanto wrong to torture others just because they have inconvenienced you.

• It is pro tanto wrong to impose severe burdens on others simply because of their physical appearance.

• It is morally admirable to express gratitude to a benefactor whose gift resulted from substantial sacrifice undertaken from exclusively altruistic motives.

• There is some moral reason to offer aid to those in distress, if such aid is very easily given and comes at very little expense.

• If acting justly is costless, then, ceteris paribus, one should act justly.

• The interests of others are sometimes morally weightier than our own.

• It is pro tanto wrong to satisfy a mild desire if this requires killing many innocent people.

We enter four important caveats at this point. First, since we are interested here only in what might qualify as a moral system for beings like us, in worlds like our own, all of the examples above ought, strictly speaking, to incorporate these qualifications. So, for instance, the first of these examples ought to read: for beings like us, in worlds like ours, it is pro tanto wrong to engage in the recreational slaughter of a fellow person. We will henceforth omit these prefatory clauses and the pro tanto qualifiers, though all subsequent examples of moral fixed points should be understood to include them.¹⁵

Second, the list of fixed points that we have provided is incomplete. And although we credit ourselves with enough conceptual mastery to confidently affirm the truth of these propositions, we acknowledge the possibility (one whose modality we discuss in Sect. 4) that we have erred in the composition of the list.

Third, and relatedly, we recognize that different moral systems will give explanatory priority to different kinds of moral concepts. Some will minimize the importance of deontic concepts, making notions of wrongness, duty, and permissibility derivative from those of evaluative, aretaic, or reason-based concepts (or, even more radically, eliminating the use of deontic concepts altogether). Others will be more comfortable placing deontic concepts at the center of their moral system. While our own formulation of the fixed points employs deontic concepts, we take no official stand on whether ours is the best or canonical formulation. Those who wish to understand the fixed points in terms of other moral categories, such as the morally better or worse, should feel free to do so. The important point is that all moral systems will employ normative concepts to express negative and positive moral assessments, and certain kinds of behavior will, as a conceptual matter, necessarily merit moral censure (whether because of their wrongness, badness, viciousness, or reason-giving force), while others will, of conceptual necessity, merit a positive moral assessment.

Finally, in our view, while the moral fixed points may or may not lie at the theoretical basis of a minimally eccentric moral system—we address this question later—they nonetheless constitute the boundaries of such a system, distinguishing it from other normative systems. Think of things this way: imagine that we could lay out the various actual and possible normative systems as if they were territories on an atlas depicting normative space. We can picture the minimally eccentric moral system as a chunk of conceptual territory on this atlas, having recognizable if vague boundaries. It is the moral fixed points that set these boundaries, distinguishing this moral system from other normative systems, such as those that govern prudential and aesthetic matters.

In presenting nonnaturalism as a position that accepts these “moral boundaries,” we realize that we are committing ourselves to considerably more than what most other nonnaturalists have explicitly endorsed. Still, our view is not without precedent. When Thomas Reid defended the claim that the “first principles of morals” are constitutive of competent moral thought, he endorsed a similar view.¹⁶ And when Philippa Foot argued against Hare’s position by calling attention to so-called thick evaluative predicates such as “rude” and “just,” whose application conditions are fixed by their descriptive content, she also defended a position similar to ours. In Foot’s view, there are certain in-built conceptual constraints on what could qualify as rude or just behavior, thereby making it the case that some actions, by definition, could not qualify as rude or just.¹⁷ Indeed, one could view the position we defend as an attempt to extend Foot’s basic insight that (to put things in our terms) there are conceptual limits as to what could count as the elements of a minimally eccentric moral system. For an implication of our view is that inasmuch as there are discernible boundaries to the application conditions of so-called thin concepts such as ‘wrong,’ these concepts behave much more like thick ones than many have believed.

Suppose, for argument’s sake, that the picture we’ve sketched is more or less correct, and that the moral fixed points set the boundaries of a minimally eccentric moral system. Still, one might wonder: Why should we care about morality? Why think that we have reason to pledge our allegiance to this normative system, rather than another—call it schmorality—that fails to incorporate (perhaps some central cases of) the moral fixed points?

It’s an excellent question, but one that we don’t propose to answer here. Two reasons explain our reticence. First, an adequate answer would take (at the least) an essay unto itself. But, second, this question is a perennial worry for all forms of moral realism. Naturalists as well as nonnaturalists face this difficulty, in equal measure. And while regarding some substantive moral norms as a species of conceptual truth might not specially aid us in explaining the reason-giving power of moral facts, neither does it make our version of realism any the less apt to offer such an explanation, whatever it may be.¹⁸

3 The nature of the truths

Is there an illuminating way to characterize nonnaturalism? We have suggested that there is. Think of the view as being committed to:

The Core Claim: there are nonnatural moral truths.

In the last section, we added some content to this abstract claim, maintaining that among the nonnatural truths are the moral fixed points, which are constitutive of any minimally eccentric moral system. Our project in this section is to both deepen and expand upon our presentation of the moral fixed points, offering an account according to which they are a species of conceptual truth.

The moral fixed points, we have said, are immensely plausible candidates for being first-order moral truths. They are also constitutive of anything worth calling a minimally eccentric moral system. To return to the metaphor on which we’ve relied, they fix the boundaries of such a moral system, enjoying, as we’ll put it, framework status. In an exchange with Ronald Dworkin, Sharon Street—who is no friend of realism—goes a step further. Street writes that it seems “almost crazy” to deny that slaughtering a young baby right before her captive mother’s eyes is wrong.¹⁹ This seems exactly right. Indeed, this strikes us as the kind of evidence one should expect for any proposition that qualifies as a species of conceptual truth.

While we have no general criterion for determining whether a proposition p is a type of conceptual truth, here are several signs that we are in the neighborhood. First, p is, if true, necessarily true. Not all necessary truths are conceptual truths, of course, but paradigmatic conceptual truths will be necessary ones. Second, p enjoys framework status, fixing the boundaries as to what counts as a type of subject matter. For example, the proposition 〈that God is a perfect being〉 arguably enjoys framework status; were you to reject the sentence “God is a perfect being” but to engage in theological discourse, this would be good reason to hold that you are not employing the theistic concept ‘God.’ Third, p’s denial would tend to evoke bewilderment among those competent with its constituent concepts—a response to the effect that its denial would be almost crazy. To deny the proposition that 〈sets are identical if and only if they have the same members〉, for example, would tend to evoke such a response, and so is a good candidate for being a conceptual truth.²⁰ And, fourth, p is knowable a priori, simply by adequately understanding its constituent concepts and their relations to one another. Both of the propositions that we have just mentioned are, in our judgment, good candidates for being knowable a priori. Their warrant derives simply from understanding concepts such as ‘God,’ ‘set,’ and ‘member’ and the relations they bear to one another.

The moral fixed points, we believe, bear all of these marks of being a species of conceptual truth. If they are true, then they are necessarily true. Moreover, they enjoy framework status, are so plausible that their denial would tend to evoke bafflement among those competent with their constituent concepts, and are good candidates for being knowable a priori if they are knowable at all.²¹ These are, as we say, “marks” of conceptual truths. Our claim is not that a proposition’s exemplifying all four marks entails that it is a species of conceptual truth, only that we have good reason to hold that it is.

We are well aware of the resistance likely to greet our claim that the moral fixed points are a species of conceptual truth. Such resistance is bound to come from both friends and foes of moral nonnaturalism. Friends of nonnaturalism will worry that the position is incompatible with the Open Question Argument, which is one of the primary motivations for their view. Foes of nonnaturalism, such as error theorists, will think it highly uncharitable (to say the least) to assign conceptual error to those who reject the moral fixed points. Still others will accept the truth of the moral fixed points, while denying their status as a type of conceptual truth. These are certainly concerns worth addressing. Before we undertake that task, however, we need first to forestall a possible misunderstanding.

Many philosophers are accustomed to thinking of conceptual truths as formal or vacuous truths that are obvious. Clearly, that is not how we are conceiving of them. We hold that some conceptual truths have substantive content and needn’t be obvious. The moral fixed points are hardly empty tautologies, but that doesn’t distinguish them from many non-moral propositions plausibly viewed as conceptual truths. That justified true belief is insufficient for knowledge; that God’s possible existence entails God’s necessary existence; that meaningful statements may be neither empirically verifiable nor analytic—these are substantive truths, surely, and yet also good candidates for being conceptual ones.

That noted, the plausibility of whether a given proposition is a type of conceptual truth depends on how one understands the nature of propositions and the nature of the concepts that compose them.²² Our task in this section is, in part, to present an account of the nature of concepts that helps to substantiate the claim that the moral fixed points are plausible candidates for being a species of conceptual truth. In this respect, two important parts of our project—defending the claim that the moral fixed points are a type of conceptual truth and the commitment to a particular account of the nature of concepts—are deeply connected.

Let us begin with concepts. The position we defend regarding the nature of the moral fixed points rests on a thoroughly traditional understanding of concepts and their contribution to thought, which runs from Aristotle through Frege. According to this understanding, which we will call the traditional view, concepts have three distinct characteristics.

In the first place, concepts are not mental or linguistic entities. They are not ideas in the head, sentences in the brain, or anything of that sort. Rather, they are abstract, sharable, mind-independent ways of thinking about objects and their properties. As such, they are very much objective, “out there” sorts of things, extra-mental items whose existence does not depend on our employing them in thought or language. Examples of concepts include not only predicative ones such as ‘being a horse,’ but also those that do not (or need not) play a predicative role, such as demonstratives (e.g., ‘that,’ ‘there’) and perhaps indexicals (e.g., ‘now,’ ‘today’).

Second, concepts are the building blocks or sub-components of propositions, which are themselves the objects of belief and other propositional attitudes. If this is right, then many propositions are constituted by concepts of objects and properties, rather than by objects and properties themselves. That said, the traditional view has the flexibility to allow that propositions come in several varieties. It might be, for example, that some propositions are constituted not entirely by concepts and relations but also by individuals. So-called singular propositions such as 〈that Obama owns a pair of glasses〉, for example, might have Obama himself as a constituent. Similarly, as philosophers such as Hilary Putnam have suggested, it might be that propositions concerning natural kinds are constituted not entirely by concepts and relations but by the kinds themselves.²³ The proposition 〈that a gallon of water weighs over eight pounds〉, for example, might have as one of its constituents not the concept ‘water’ but the natural kind water. In what follows, we prefer to remain neutral on whether there are propositions with constituents such as these; the important point here is that the traditional view sees concepts as at least the partial, and sometimes (in addition to relations) the exhaustive, constituents of propositions, which are themselves objects of thought.²⁴

Third, according to the traditional view, concepts are not merely the subcomponents of propositions; they are also referential devices or ways of getting things in mind that enable thinkers to refer to things such as objects and properties. Although concepts need not apply to actual objects, properties, or the like—as there may be none that fall under them—their function is such that they purport to apply to them. Indeed, in a range of paradigm cases, concepts necessarily determine the range of entities that they are about.²⁵ In this regard, concepts appear to be very different from linguistic entities such as words. The concept ‘being wrong,’ for example, could not be the concept it is if it were not about wrongness; it belongs to the essence of the concept that it applies to exactly those things that are wrong (if any such things there be). The word “wrong,” by contrast, could be the word it is even if it were not about wrongness.²⁶ What determines whether this term applies to all and only the wrong things? If the traditional view is correct, it is the fact that the term expresses the concept ‘being wrong.’

This last point about concepts is particularly important for our purposes, for it highlights the fact that concepts have essences. It belongs to the essence of the concept ‘being wrong,’ for example, that it applies to exactly those things that are wrong. And once we see that concepts have essences, we are in a position to say what it is for a proposition to be a conceptual truth.

According to the traditional view, a proposition 〈that x is F〉 is a conceptual truth if it belongs to the essence of ‘F’ that, necessarily, anything that satisfies ‘x’ also satisfies ‘F.’ To see how this works for the moral fixed points, consider the proposition 〈that recreational slaughter of a fellow person is wrong〉. This is a conceptual truth in case it belongs to the essence of the concept ‘being wrong’ that, necessarily, if anything satisfies the concept ‘recreational slaughter’ (of a fellow person) it also satisfies ‘being wrong’ (in a world sufficiently similar to ours).²⁷

Thinking about conceptual truths in this way has several interesting implications. One is that we should distinguish conceptual from analytic truths. Conceptual truths are true propositions; specifically, those that hold in virtue of the essences of their constituent concepts. By contrast, analytic truths (at least under one historically prominent understanding) are sentences: those that are true solely in virtue of the meaning of their terms. As we hinted above, advocates of the traditional view typically claim that concepts are the meanings expressed by our words. Still, not all concepts are expressed linguistically. Some we have yet to discover; others may permanently elude us, owing, perhaps, to their complexity. Two things follow: first, conceptual truths are not analytic truths, since the former are propositions while the latter are not. And, second, not all conceptual truths are expressed by analytic truths, even though it might be that all analytic truths express conceptual truths.

Another implication of the traditional view concerns the relations between propositions and facts. There is a familiar view—call it the correspondence view—according to which propositions and facts belong to different categories. As the correspondence view has it, propositions are representational entities; they represent what is the case. Facts, by contrast, are not representational entities, but are that which is represented. If this familiar view is correct, facts are not identical with true propositions; rather, facts correspond to true propositions, making them true.

In its main lines, we find the correspondence view highly attractive. It is quite plausible to hold that a proposition—say, 〈that the current U.S. President owns a pair of glasses〉—is true just because it is a fact that Barack Obama, the current U.S. president, actually owns a pair of glasses. Indeed, in many ways, the traditional view of concepts fits nicely with the correspondence view. For if concepts are both devices of representation and the sub-components of propositions, this would help to explain why propositions are by their very nature representational, representing facts of various sorts.

Still, if the traditional view of concepts is correct, the correspondence view does not offer a fully general account of truth. For conceptual truths do not require facts that correspond to them, worldly truth-makers that render them true. Whether such propositions are true depends on the natures of their constituent concepts and not solely on the existence of any correlative worldly facts that can serve as truth-makers.²⁸ (By “worldly facts,” we simply mean facts that hold but not solely in virtue of the essences of concepts.)

To see this, suppose that the following moral proposition is a conceptual truth: 〈that it is wrong to impose agony on others solely for personal gain〉. The truth of this proposition is compatible with there being a worldly fact that it is wrong to impose agony on others solely for personal gain. Still, this fact does not itself make the proposition in question true. The proposition is true in virtue of the essence of its constituent concepts—these concepts being such that their having the nature they do does not depend on there being some property or fact to which they refer.

Given these points, nonnaturalists have at hand the resources to move beyond their Core Claim and to advance:

The Embellished Core Claim: There are nonnatural moral truths. These truths include the moral fixed points, which are a species of conceptual truth, as they are propositions that are true in virtue of the essences of their constituent concepts.

The Embellished Core Claim is an implication of combining the traditional view of concepts with the thesis that the moral fixed points are conceptual truths composed of nonnatural moral concepts. For suppose we assume that the moral fixed points are true of conceptual necessity. And suppose we also assume that these truths are constituted by moral concepts, which are nonnatural. The Embellished Core Claim follows straight away. The moral fixed points are true in virtue of their constituent concepts, some of which are nonnatural.²⁹

4 Epistemic implications

The Embellished Core Claim is an ontological thesis, telling us something about the nature of a range of moral truths. In the next section, we will unpack this thesis in greater detail, contending that it renders nonnaturalism a highly attractive position for realists of all sorts to accept. For now, however, let us highlight some of the view’s epistemic implications.

To that end, consider a figure whom we will call the nihilist. The nihilist does not have eccentric empirical beliefs, such as the belief that we enjoy pain. Yet he sincerely denies that any of the moral fixed points are true. The nihilist, it should be added, needn’t be a moral monster. He may be deeply averse to anyone’s performing horrific acts, such as (to advert to Street’s earlier example) killing a small child in his captive mother’s presence. He might also acknowledge that killing a child in this way seems to be wrong, even on reflection. Still, the nihilist denies that were you to commit such a horrific act, you would have thereby done anything immoral.

What should nonnaturalists say about the nihilist? If the moral fixed points are a species of conceptual truth, then nonnaturalists should say that the nihilist suffers from a conceptual deficiency. Such a deficiency would, presumably, consist in the nihilist’s being such that he either (i) lacks the relevant concepts, such as the concept ‘being wrong’; (ii) has an inadequate grasp of them, failing to see that they apply to some paradigm cases; (iii) has a confused grasp of them, perhaps holding that the concept ‘being wrong’ fails to apply to acts taken toward those who belong to some “out group,” such as those conquered in battle; or (iv) fails to appreciate or acknowledge the manifest implications of the concepts in question, owing to some aspect of his psychology or situation that inhibits him from acknowledging that the killing is wrong (even though it may seem to him to be wrong). In failures of this last sort, the lack of appreciation or acknowledgment might be because the nihilist is deeply impressed by Mackie’s arguments from queerness that there are no moral truths. Or perhaps it is because he is baffled by the idea that the world that science describes could also include values not of our own creation. Or, more darkly, it might be because he is like the Nazi war criminal Herman Göring, who justified the slaughter he ordered by appealing to the ardent love of his people.³⁰

As should be evident, we are operating with a fairly capacious account of what it is to suffer from a conceptual deficiency. It is worth stressing two points about this account. In the first place, in saying that the nihilist suffers from a conceptual deficiency, we do not thereby commit ourselves to the claim that she suffers only from a conceptual deficiency. She may suffer from a very serious moral deficiency as well. Nor do we commit ourselves to the claim that, in the nihilist’s case, the conceptual deficiency is somehow fundamental. Here, as elsewhere, there may be multiple factors at work that explain why it is that someone rejects the moral fixed points.

In the second place, our view does not imply that if someone were to suffer from a conceptual deficiency, she is thereby conceptually confused. That is, our view does not imply that such a person applies moral concepts incorrectly, makes arbitrary discriminations in what she applies them to, or has seriously confused thoughts that she associates with these concepts. For, as we think of it, a conceptual deficiency needn’t involve conceptual confusion in the sense just explained; it can simply consist in failing to appreciate or acknowledge the manifest implications of a concept or its application to a particular case. Otherwise put, in some cases, a conceptual deficiency can consist not simply in applying a concept incorrectly but in failing to see or acknowledge that it has application, where this should be manifest to an ordinary agent.

We can fill out our diagnosis of the nihilist’s conceptual errors in more detail. Nonnaturalism of the variety we advocate implies that:

If the moral fixed points are true, then they are true of conceptual necessity.

That is, if we hold certain descriptive information fixed—such as our present human constitution and environment—the concept ‘being wrong’ is such that it belongs to its essence that, necessarily, if anything falls under the concept ‘recreational slaughter’ (of a fellow person), then it also falls under it. As we’ve already emphasized, not all conceptual truths are obviously true; they can be evident to different degrees. The proposition that all red things are red is obviously true; the proposition that justified true belief is insufficient for knowledge was (pre-1963) anything but.

Nonnaturalists hold that the degree to which the moral fixed points are evident is quite high.³¹ While carefully reflecting on the proposition 〈that torturing others just for fun is wrong〉, it is very difficult to conceive of a situation in which there is nothing morally amiss with such behavior. (Indeed, some nihilists, such as Bart Streumer, maintain that their view is so radical as to be literally incredible.³²) But nonnaturalists do not maintain that the moral fixed points are maximally evident. After all, when not carefully attending to the proposition 〈that torturing others just for fun is wrong〉, it is possible to wonder whether the proposition is true. It is possible to wonder, for example, whether one has failed to appreciate the force of various antirealist arguments, such as Mackie’s arguments from queerness.³³ It is also possible to wonder whether there is something deeply defective about our moral concepts, which we have not yet appreciated, which would render them incapable of referring to moral properties if any were to exist. In short, our position does not imply that it is incoherent to deny a conceptual truth.

The dynamic of doubt that we have just described is familiar in philosophy. While attending to the proposition that, necessarily, 2 + 3 = 5, it is very difficult to see how there could be no numbers. But then one steps back and wonders whether Hartry Field’s arguments for numerical nihilism might have bite. While attending to the law of noncontradiction, one finds it impossible to see how there could be true contradictions. But then one is introduced to the Liar Paradox and becomes open to the possibility that dialethism might be correct. In each of these cases, one acknowledges that although a proposition seems highly evident, it might nonetheless be false.

What does this “might” come to? It is not easy to say. It is not a species of broadly logical possibility. Nonnaturalists cannot coherently say that there is some possible world like ours, with beings such as us, in which the moral fixed points are false. Nor is the “might” a species of epistemic possibility, if we mean by this that, for all we reasonably believe, there is a world like ours with beings such as us, in which the moral fixed are false, since nonnaturalists deny that our reasonable beliefs license such a possibility.

The best that can be said is probably this. All of us are aware of cases in which a proposition has seemed evidently true, even on extended reflection, but has turned out to be false. Call such propositions illusory. Nonnaturalists can agree that, given our evidence, we cannot rule out that the moral fixed points belong to the class of illusory propositions. It is in this sense—which, admittedly, is fairly weak—that nonnaturalists maintain that we might be wrong about the status of the moral fixed points. That a proposition might be illusory is not, however, sufficient reason to reject its status as a conceptual truth.

5 How deep the divide?

Let us pull the various strands of our discussion together. Nonnaturalism, we have suggested, is best understood to incorporate:

The Core Claim: there are nonnatural moral truths.

Suppose, though, that among the moral truths are those that we have called the moral fixed points: immensely plausible first-order moral propositions such as 〈that it is wrong to engage in recreational slaughter of a fellow person〉 or 〈that it is wrong to rape a child solely to indulge one’s lust〉. Suppose, too, that we accept the traditional view of concepts, according to which many propositions are constituted by concepts, maintaining that conceptual truths hold in virtue of the essences of their constituent concepts. Given that moral concepts are nonnatural and that the moral fixed points are a species of conceptual truth, moral nonnaturalists are well-positioned to accept a more ambitious thesis, which we have labeled:

The Embellished Core Claim: There are nonnatural moral truths. These truths include the moral fixed points, which are a species of conceptual truth, as they are propositions that are true in virtue of the essences of their constituent concepts.

In this section, we recommend The Embellished Core Claim to both moral naturalists and our fellow nonnaturalists. We stress that our recommendation is not supposed to be an argument that other moral realists ought to accept nonnaturalism of the sort we endorse. That would require a defense of the traditional view of concepts, which is outside our remit. To the extent that we provide an argument for The Embellished Core Claim, it comes in the next two sections and is indirect. We think that realists have reason to accept it mostly because it does a lot of explanatory work, providing the resources to respond in an elegant way to a wide array of the most worrisome objections pressed against their views.

We recognize that many moral realists will not be immediately attracted to The Embellished Core Claim. Some naturalists will think that accepting it is too much, as it fits very uneasily with some of their central naturalistic commitments. Some nonnaturalists, by contrast, will hold that accepting it is not enough. These philosophers will insist that we should commit ourselves not only to nonnatural moral truths, but also to nonnaturalistic moral properties and facts. In what follows, we try to allay some of these concerns.

We begin by addressing the naturalists. Naturalists have tended to be fairly dismissive of nonnaturalism. Allan Gibbard, for example, calls the view “mumbo jumbo.”³⁴ Frank Jackson says that it is “spooky.”³⁵ Characterizations such as these notwithstanding, nonnaturalism of the sort we defend is not driven by extravagant ontological claims. It is best understood to be the implication of combining the thesis that moral concepts are nonnatural with a fairly traditional account of the nature of concepts and conceptual truths according to which these concepts are the components of the moral fixed points. These commitments are, as far as we can tell, not at odds with naturalism. As we have indicated already, it is difficult to find naturalists who explicitly reject the claim that moral concepts are nonnatural. Moreover, the traditional view of concepts is perfectly compatible with what many naturalists say about the nature of moral thought.

This last claim might strike some as surprising. But it shouldn’t, as the traditional view of concepts is compatible with a variety of views about what explains how concepts refer. Suppose, for example, that one is a descriptivist about reference, as are the so-called Canberra naturalists such as Frank Jackson, Philip Pettit, and Michael Smith. The traditional view of concepts can be combined with (although it does not entail) such a view. It may be, for example, that the concept ‘being wrong’ refers of necessity to all and only wrong actions in virtue of its presence in a system of platitudes that uniquely determines the reference of the term “wrong.”³⁶ Or suppose that one is a Cornell realist, such as Richard Boyd, holding that the term “being wrong” applies exactly to wrong actions in virtue of the fact that its use is causally regulated by the property being wrong.³⁷ The traditional view of concepts is compatible with this position as well. Were Boyd to countenance the existence of concepts, then his view would (like that of the Canberra naturalists) simply provide a particular kind of explanation as to why the concept ‘being wrong’ applies to all and only wrong actions. As such, these naturalistic views are fully compatible with the more general theses about concepts that we favor.

To be clear: we mean to endorse neither of these naturalistic hypotheses. In fact, we believe that both face serious obstacles. Rather, our point is that it is important to distinguish the question of what concepts are from the different question of what explains how they refer. Naturalists have said relatively little about the first question, focusing on the second (although often couching their views as applying to terms rather than concepts). So far as we can tell, however, accepting The Embellished Core Claim is compatible with a good deal of what naturalists have said about the nature of reference.

There is, however, a more important point in the offing. At the core of any naturalist realist project is the ontological thesis that there are moral properties and facts, which are natural. Most have thought this thesis to be incompatible with anything worth calling moral nonnaturalism. But if the view we’ve defended is correct, this is not so. To see this, return to our earlier distinction between:

Minimal nonnaturalism: there are nonnatural moral truths, but there are no nonnatural moral properties or facts. All moral properties and facts are natural.

and:

Robust nonnaturalism: there are both nonnatural moral truths and nonnatural moral properties and facts.

Minimal nonnaturalism is not only compatible with the naturalists’ core ontological commitments, but also gets naturalists a good deal of what they want from a metaethical view. For it provides the resources to accommodate some of the criticisms that nonnaturalists have pressed against naturalism, including the worry that moral thought is generally in the business not of predicting or causally explaining the sorts of things that are the subjects of the natural sciences, but rather of evaluating and directing behavior.³⁸ Because minimal nonnaturalism understands moral propositions to be constituted by irreducibly moral concepts, which are both evaluative and action-guiding, the view appears well-positioned to accommodate the evaluative and action-guiding character of moral thought. We believe that this is a reason for naturalists to take Minimal nonnaturalism very seriously.

We now turn to our fellow nonnaturalists. In our view, accepting The Embellished Core Claim also gets nonnaturalists much of what they want from a metaethical view. After all, if the Embellished Core Claim is correct, then the moral fixed points are true not exclusively in virtue of there being natural facts that make them true. Rather, they are true in virtue of the essences of their constituent concepts, which are nonnatural. This is a fairly robust ontological thesis. It may be robust enough that, when it comes to views about the nature of moral properties and facts, nonnaturalists can be noncommittal, letting the chips fall where they may.

Still, we suspect that many nonnaturalists will want more. In the remainder of this section, we contend that, given a few additional assumptions, The Embellished Core Claim implies something very much in the spirit of Robust nonnaturalism. We call this line of reasoning the Reversal Argument.

The Reversal Argument concerns the relationship that truths bear to facts, reversing the standard order of explanation between them. That order is captured by the claim that, necessarily, every truth requires a truth-maker, which is a correlative worldly fact. Earlier we claimed that we should reject this thesis, as some truths do not require independent correlative worldly truth-makers. These are the conceptual truths, which are true in virtue of the essences of their constituent concepts. The Reversal Argument proposes putting some of these truths to work. Conceptual truths do not require correlative truth-makers; rather, some are truths that are fact-makers. The moral fixed points, according to the argument, are truths of this sort.³⁹

When stated explicitly, the Reversal Argument runs as follows:

• (1) There are particulars, such as act-tokens and individual people, which have moral properties. A particular’s having a moral property is a particular moral fact.

• (2) When a particular has a moral property, it has that property in virtue of there being some moral standard which that particular satisfies or fails to satisfy.

• (3) Some of these moral standards are the moral fixed points, which are conceptual truths.

• (4) So, the moral fixed points are fact-makers: there are some particulars that have the moral properties they do in virtue of the moral fixed points.

Let’s take a closer look at the argument’s three premises.

All moral realists, to our knowledge, accept the argument’s first premise, even if they may be reticent about stating how exactly they think of moral facts. By contrast, not all moral realists accept the argument’s second premise—particularists, who reject the existence (or usefulness) of general moral standards, being the notable exception.⁴⁰ For present purposes, we propose to bracket particularism, suggesting that realists who reject this view—and we count ourselves among them—should accept the argument’s second premise. In an important defense of nonnaturalism, William FitzPatrick offers an argument to explain this commitment.⁴¹

Consider, says FitzPatrick, an artifact such as the computer on your desk. Now ask yourself why that computer is a good one. Your answer will appeal to ordinary natural properties such as its processing speed, its propensity not to crash, and so forth. Following FitzPatrick, call these features of your computer XYZ. Although your computer is good in virtue of having these properties, its being good does not consist solely in its having them, for any number of things could have these properties without being good. When it comes to artifacts, at least, good-making characteristics are relative to kinds; sharpness is a good-making quality in a knife, for example, but not in a coffee cup. FitzPatrick maintains that this point is important

not because …[it]… translates directly to the ethical case—it does not, as at least some ethically good- or bad-making properties may plausibly have that status invariantly … but because it directs us to the more complex structure of facts about goodness, which does have an analog in ethics. In the present case, the point is that the fact that this computer is good consists not only in the fact that it has XYZ, but in this together with the fact that XYZ is such as to satisfy the standards of goodness S for computers.⁴²

By noting that the goodness of an artifact of some kind is determined by the standards of goodness for members of that kind, FitzPatrick does not wish to defend nonnaturalism about such standards. Rather, he wishes to direct our attention to, as he puts it, the structure of goodness-involving facts, or as we would prefer to put things, the structure of normative facts.

FitzPatrick’s argument is one from parallel cases: to account for the normative properties of artifacts, we must appeal to normative standards, such as the standards of goodness for computers or knives, which explain how their natural properties contribute to their goodness or badness qua artifact. Likewise, to account for the moral properties of actions, we need to appeal to moral standards, which explain how an action’s natural properties contribute to its having such properties as being morally permissible or impermissible.

To illustrate the point, imagine being present at a social gathering in which a senior colleague, who has a strong sadistic streak, publically humiliates a junior colleague with a clever but vicious put-down. Your colleague’s behavior is morally wrong. But it is wrong not simply because it has a constellation of ordinary natural properties, e.g., its being such as to cause humiliation and social tension. It is also wrong because there is the moral standard that:

It is wrong to humiliate others simply for pleasure,

which applies to your colleague’s behavior. We would add that this argument from parallel cases can be extended to incorporate other normative domains. For the parallel that holds is not simply between the structure of normative facts involving artifacts, on the one hand, and the structure of moral facts, on the other. It is also between the structure of moral facts and other normative facts that involve practical activity.

Imagine—to stay with our example—that your colleague’s behavior is not merely wrong but also rude and practically irrational. If his action is rude, however, then it is in virtue of his having violated standards of politeness that apply to his behavior. And if his action is practically irrational—say, because it deeply alienates those whose company he generally enjoys—then it is in virtue of his having violated applicable standards of prudence. In short, the structure of normative facts that FitzPatrick has explicated is pervasive; it is found not simply in the moral domain but in normative domains quite generally.⁴³

The final premise of the Reversal Argument is that, when an action has some moral property such as being wrong, it has it in virtue of its satisfying or failing to satisfy a moral standard, some of which are moral fixed points.⁴⁴ If moral standards are identical with propositions, then securing this last premise is a straightforward matter. After all, if a moral standard accounts for why a particular case of recreational slaughter is wrong and this standard is a fixed point, then a moral fixed point explains why this killing is wrong. That said, one might hold that for every moral fixed point there is correlative worldly fact, even though that fact is not required to make that fixed point true. If one accepted a view such as this, then we would need a reason to believe that it is the fixed points rather than (simply) their correlative worldly facts that are serving as the standards that play the fact-making role. We believe that there is such a reason; we offer it here as an argument from ontological economy.

Suppose, for argument’s sake, that The Embellished Core Claim is true: there are a variety of moral fixed points, each of which is true of conceptual necessity. On the face of things, these truths appear to be ideally suited to explain why particulars have moral properties, as such truths are the sort of thing that ensure, by their very essence, that if something is a case of (say) recreational slaughter, then it has the property being wrong. Under this reading, it belongs to the nature of these propositions to bind together the non-moral and the moral, ensuring, for example, that any instance of recreational slaughter of fellow persons is wrong.

Now suppose that for every such conceptual truth T, there is a worldly fact that corresponds to it and is constituted by the properties that the concepts constitutive of T express. If this view were correct, then there is a worldly fact that recreational slaughter of a fellow person is wrong which corresponds to the proposition 〈that recreational slaughter of a fellow person is wrong〉. Notice, however, that even if there were such a fact and it were constituted by the property being wrong, it needn’t belong to the essence of this property that it explains why recreational slaughter of a fellow person must be wrong. Of course one could understand wrongness in such a way that it does explain this. But in our view, there would be no need to, as it already belongs to the essence of the concept ‘being wrong’ to explain why any case of recreational slaughter of a fellow person must be wrong. So, on grounds of parsimony—best not to invoke two things to do the needed explaining, when one will suffice—we should hold that the moral fixed points are fact-makers.

Suppose that the argument we’ve offered hits its mark: the moral fixed points are that in virtue of which some particulars have moral properties—“fact-makers,” as we’ve called them. Does this imply that Robust nonnaturalism is true? Not exactly. Robust nonnaturalism says that moral properties and facts are nonnatural. The Reversal Argument, however, establishes only that a certain class of moral fact-makers is nonnatural. It would imply Robust nonnaturalism only if moral fact-makers are also constituents of moral facts. Whether they are, however, depends on how facts are constituted.

On this matter, we see two options. According to a wide account of moral facts, the fact that your senior colleague’s behavior is wrong must include within it all that makes his action wrong, including the fact that he has insulted your junior colleague, that he has done so for the mere pleasure of it, and so forth. According to a narrow account of moral facts, there is a principled distinction to draw between what makes your colleague’s action wrong, on the one hand, and the fact that your colleague’s behavior is wrong, on the other—however difficult it may be to specify this distinction exactly.

If the wide view is correct, it follows that there are nonnatural moral facts and, hence, that something close to Robust nonnaturalism is true. For the fact that your colleague’s behavior is wrong must incorporate everything that makes his action wrong, including the standard:

that it is wrong to humiliate others simply for pleasure,

which, according to our view, is a nonnatural truth.⁴⁵ The narrow view of facts, by contrast, has more modest implications, since it distinguishes moral fact-makers from the moral facts themselves. If the narrow view were correct, then, for all we have said, every moral fact (and every moral property) may be a natural one, even if the standards that account for the existence of these particular moral facts—the fact-makers—are nonnatural.

Earlier we indicated that the Reversal Argument yields a position in the spirit but perhaps not the letter of Robust nonnaturalism. The explanation of that caveat should now be clear. Whether the argument implies Robust nonnaturalism would depend on (among other things) resolving issues about how to individuate facts, which we have no intention of trying to do. In our view, however, nonnaturalists shouldn’t be too exercised about this issue. They should be comfortable with either of the nonnaturalist positions we’ve sketched. For according to each, nonnatural moral truths exist and are among the fact-makers of the world, whether or not they are the constituents of particular moral facts. Once these points are secured, the remaining differences between Minimal and Robust nonnaturalism strike us as matters of only secondary importance.

6 Payoffs

Our project has been to develop a robust version of nonnaturalism that is explicit about its ontological commitments and attractive to naturalists and fellow nonnaturalists alike. But why accept it? Earlier we said that the primary argument for accepting this view would be indirect, consisting in showcasing its explanatory power. In this section, we present this argument, highlighting how reliance on the moral fixed points can provide nonnaturalists with the resources to respond to prominent objections to their view. We focus on three such objections.

(1) The first objection: Moral Disagreement

To what extent should our confidence in the truth of our moral beliefs be rattled by deep and persistent moral disagreement? Those who advance arguments from moral disagreement maintain that our confidence should be shaken to the point that we surrender these beliefs. While arguments from moral disagreement come in a variety of forms, we can sort them initially into two broad categories: those that are supposed to have metaphysical implications and those that are supposed to have only epistemological ones. Arguments of the first sort allege that the best explanation of moral disagreement implies that there are (probably) no moral truths, realistically construed.⁴⁶ Arguments of the second sort attempt to show that even if there are such truths, the breadth and depth of moral disagreement defeats any warrant that our moral beliefs might otherwise enjoy.⁴⁷

The disagreement in question can be of two kinds. On the one hand, the parties to the disagreement may be actual agents. On the other hand, they may be idealized agents in idealized circumstances of reflection and choice. Regardless of which sort of disagreement is at issue, it is essential to note that the disagreement that calls for explanation must obtain among sufficiently competent interlocutors. After all, if some people are much better situated than others to assess a given matter, then the existence of dissenting voices offers no support for an anti-realist diagnosis of the issue that divides them, and should not incline us to pessimism about the prospects for knowledge. Tens of millions of people disagree with the scientific consensus that the Earth is more than 7,000 years old. This hardly forces us to an anti-realist construal of geological or cosmological claims. Nor does it defeat the warranted beliefs of those who judge the Earth to be many billions of years old.

Keeping in mind that the relevant sort of disagreement must be that which obtains between those who are competent to judge the matters at hand, consider once again moral fixed points such as:

〈that it is wrong to engage in the recreational slaughter of a fellow person〉

〈that it is wrong, simply for convenience’s sake, to break promises on which others are relying〉

And:

〈that it is wrong to torture others just because they have inconvenienced you〉.

The first thing to note is that we do not in fact see widespread actual disagreement about moral fixed points such as these among fully and equally conceptually competent agents.⁴⁸ To the contrary, the fixed points have gained near-universal endorsement across times and cultures. Indeed, if what we have said is correct, disagreement about the truth of these propositions among fully conceptually competent agents would be impossible, since any case in which a moral disagreement were the result of an agent’s rejecting the moral fixed points would be one in which that agent suffers from a conceptual deficiency, and hence, is not a fully competent interlocutor.

The second thing to note is that, if this is so, then we should not anticipate hypothetical disagreement about the fixed points among agents who are both fully competent to judge moral matters and positioned in ideal circumstances of assessment. That is because their competence implies, among other things, a disposition to sincerely affirm the moral fixed points when contemplating their content.

The upshot is that at least a substantial core of moral beliefs is immune to metaphysical and epistemological worries that stem from moral disagreement. Admittedly, this defense does not itself protect moral beliefs whose contents are not fixed points from the arguments from disagreement. Still, to the extent that considerations of disagreement are meant to undermine nonnaturalist realism per se, or the existence of any moral knowledge on nonnaturalist assumptions, then reliance on the moral fixed points can show that such ambitious arguments are mistaken. Once we insulate at least some substantial moral truths and some moral knowledge from arguments stemming from disagreement, it becomes an open question how much more truth there is, and how much additional knowledge we can gain by standard inferential means from knowledge of the moral fixed points.

Still, numerous philosophers have assumed that since we should expect to see convergence over time in domains that are best construed realistically, nonnaturalists owe an explanation of why some people steadfastly refuse to sign on to apparently obvious moral truths such as the fixed points.⁴⁹ But our view identifies such an explanation, as it implies that such a refusal is rooted in one or another conceptual deficiency. These deficiencies, in turn, will either themselves consist in or be accompanied by familiar kinds of doxastic maladies. In many cases, the deficiencies are the cumulative effect of bias, prejudice, various kinds of irrationality, factual ignorance, or limited imagination.⁵⁰ In other cases, such as that of the nihilist discussed earlier, the deficiency will consist in an agent’s having been persuaded by unsound anti-realist arguments that deploy premises that are considerably less evident and less plausible than are the moral fixed points.⁵¹

Of course nihilists may well bristle at such a characterization—they will regard the arguments that have persuaded them as sound ones, and will deny any conceptual shortcoming when it comes to moral matters. We don’t claim to have shown that their response is mistaken; we have not, after all, defended the traditional view of concepts, and have done nothing to engage directly with many of the arguments in the anti-realist’s arsenal.⁵² Still, if we are right, then nihilists, and others who reject the moral fixed points, are not our epistemic peers when it comes to moral matters. (Since competence is relative to areas of inquiry, they may well be our epistemic peers or superiors with regard to many other matters.) The mere fact, then, that these people reject the moral fixed points does not force us to acknowledge that there is moral disagreement of such a kind that yields either moral antirealism or moral skepticism.

(2) The second objection: Remarkable Coincidence Arguments

Consider an uncertain but not implausible empirical hypothesis about the origins of our moral beliefs: they have arisen as a result of doxastic mechanisms and processes that are themselves the vestiges of evolutionary pressures. The forces of natural selection have inclined human beings, like members of all species, to develop behavioral dispositions that are fitness-enhancing. Among such behavioral dispositions are those that push us to form adaptive beliefs. Having normative and evaluative beliefs that counsel adaptive behaviors is precisely what we would expect were evolutionary pressures heavily at work in the shaping of our moral faculties.

Such expectations are not disappointed. We see a widespread endorsement of practices and policies that favor giving priority to family members over strangers, praising those who keep their word, condemning those who deliberately harm others, and favorably evaluating those who reciprocate kindnesses. We see a near-universal rejection of positions that oppose such attitudes.⁵³

Some moral antirealists—the “debunkers,” as we’ll call them—argue that the best explanation of this broad consensus is that evolutionary forces, which have operated more or less uniformly across our species, have shaped our moral faculties to such an extent that our network of moral beliefs is “thoroughly saturated” with their influence.⁵⁴ If the debunkers are right, our moral faculties have been caused to issue the beliefs they do as a result of pressures that select not for true beliefs but for adaptive ones. But when it comes to our moral beliefs, why think that the true and the useful coincide? The worry, of course, is that if nonnaturalism is true, then they don’t.⁵⁵

The explanation for this pessimism rests on two claims that debunkers associate with nonnaturalism. First, nonnaturalists hold that there is a uniquely correct moral system, or perhaps a very small set of such systems (we will omit this qualification in what follows), that is constituted by the fundamental moral truths, these truths being stance-independent in the sense of not being determined by our actual or counterfactual attitudes. If this assumption is correct, the truth of propositions such as 〈that the recreational slaughter of a fellow person is wrong〉 is not fixed by the actual or possible attitudes we have (or would have) toward recreational slaughter. These truths hold independently of such attitudes.

In the second place, nonnaturalists—or so the debunkers insist—agree that there are no conceptual constraints on what counts as a moral system. Sharon Street explicitly endorses this idea when she writes:

According to the normative realist, there are normative truths that hold independently of all of our evaluative attitudes. Moreover, as a purely conceptual matter, these independent normative truths might be anything. In other words, for all our bare normative concepts tell us, survival might be bad, our children’s lives might be worthless, and the fact that someone has helped us might be a reason to hurt that person in return. Of course we think that these claims are false—perhaps even necessarily false—but the point is that if they are false, it is not our bare normative concepts that tell us so.⁵⁶

If Street is right, then, as a conceptual matter, the reference of our moral concepts is wholly elastic in the sense that while they must pick out moral duties, moral virtues, or moral goods, just about anything could qualify as a moral duty, a moral virtue, or a moral good.

With these claims in hand, the debunkers present:

The Remarkable Coincidence Argument

• 1. There are no contentful conceptual constraints on what can count as a moral norm or a moral system; there is an indefinitely large set of incompatible moral systems, each of which, as a matter of conceptual possibility, may be true.

• 2. Evolutionary forces have caused us to endorse only a small subset of all such systems.

• 3. If moral nonnaturalism is true, then there is a uniquely correct moral system of stance-independent moral truths.

• 4. Given the vast range of conceptually possible moral systems, the odds that evolutionary forces have pushed us to endorse the uniquely correct moral system of stance-independent moral truths are extremely low.

• 5. Such odds entail that if moral nonnaturalism is true, then it would be a remarkable coincidence were our moral beliefs largely on target.⁵⁷

This argument is compatible with our doxastic dispositions being such that they occasionally direct us to the moral truth. But if they did, the argument implies that this would be wholly accidental. For when it comes to our moral beliefs, any correlation between the true and the useful would be a massive stroke of luck. If it is, however, then whatever presumptive justification or warrant our moral beliefs enjoy would be defeated. And that defeat implies wholesale moral skepticism. So, if moral nonnaturalism is true, and if the empirical hypothesis about the evolutionary origins of our moral faculties is correct, then we can have no moral knowledge. That is a conclusion that no nonnaturalist will accept.

There are, we believe, a variety of promising replies to this type of argument, some of which we have developed elsewhere.⁵⁸ On this occasion, we propose to rush past these replies, noting that our favored version of nonnaturalism furnishes an economical response to the Remarkable Coincidence Argument: the argument’s first premise is false. There are contentful conceptual constraints on what can count as a moral system (for beings like us in a world such as ours), and the moral fixed points provide them. Since any such system must of conceptual necessity incorporate the moral fixed points, they are guaranteed to be elements of the uniquely correct minimally eccentric moral system.

To be sure, it would be folly to insist that none of our deeply held moral beliefs are the vestiges of distorting evolutionary influences—or, for that matter, erroneous influences stemming from our upbringing, class, historical and cultural location, or the like. But it is not as if we are completely in the dark as to which of our moral beliefs are defective, for by relying on the moral fixed points, we can correct for at least some of these malign influences, subjecting the rest of our moral beliefs to critical scrutiny, certifying some, and rejecting others.

In principle, however, debunkers could retrench, asking: What are the chances that the propositions that we take to be a species of conceptual moral truth really are conceptual truths? Mightn’t evolutionary pressures be operating here as well, inclining us to regard as conceptual truths propositions that are anything but?

Yes, they might. But if debunkers were to push this line of criticism, their position would be much less plausible than that advanced in their original critique. The original critique, recall, relies on the empirical hypothesis that:

The best explanation of the very broad consensus on many moral beliefs is that evolutionary forces, which select not for true beliefs but for adaptive ones, have shaped our moral faculties to such an extent that our network of moral beliefs is thoroughly saturated with their influence.

It is not at all evident, however, that we have any reason to accept the parallel empirical hypothesis that:

The best explanation of our tendencies to classify sets of propositions as conceptual truths is that evolutionary forces, which select not for true beliefs but for adaptive ones, have shaped our rational faculties to such an extent that our network of beliefs regarding what counts as a conceptual truth is thoroughly saturated with their influence.

To the contrary, there are a lot of reasons to be skeptical of this last claim. For one thing, beliefs to the effect that certain propositions are conceptual truths are sufficiently abstract that they are likely to be at many removes from susceptibility to the pressures of natural selection. Moreover, while our abilities to understand conceptual truths and to appreciate their modal status are neither especially adaptive nor maladaptive, such abilities are the natural extension of more general powers of reasoning that are themselves surely fitness-enhancing. In this respect, these advanced cognitive skills are just like those that enable us to excel at quantum physics or set theory. Our appreciation of string theory or Russell’s paradox, after all, isn’t likely to leave us with more offspring than those who have never heard of such things. Ditto for our ability to classify propositions as a species of conceptual truth. Still, we can have reasonable confidence in such abilities, as they have proven generally reliable across a range of contexts. If our tendency to classify some moral propositions as a species of conceptual truth is simply the manifestation of such general capacities, then it is difficult to see why we should believe that it is untrustworthy, at least absent further argument.⁵⁹

In sum: The Remarkable Coincidence argument hinges on the claim that, given the vast range of conceptually possible moral systems, the odds that evolutionary forces have pushed us to endorse the uniquely correct moral system of stance-independent moral truths are extremely low. Nonnaturalists respond that there is no remarkable coincidence, for there is not a vast range of conceptually possible moral systems. Anything that counts as a moral system (for beings like us in a world such as ours) would have to include the moral fixed points.

(3) The third objection: the Humean Challenge

We now turn to a third objection to nonnaturalism. It is one whose governing idea has been with us for some time, but has been crisply and forcefully presented by Tristram McPherson in a recent essay.⁶⁰ The criticism—which we’ll call the Humean Challenge—relies on the following three claims:

SUPERVENIENCE: No metaphysically possible world that is identical to a second world in all base respects can be different from the second world in its ethical respects.

BRUTE CONNECTION: The nonnaturalist must take the supervenience of the ethical properties on the base properties to involve a brute necessary connection between discontinuous properties.

And:

MODEST HUMEAN: Commitment to brute necessary connections between discontinuous properties counts significantly against a view.⁶¹

By saying that properties A and B are discontinuous, McPherson means that A and B are neither reducible to one another, identical to one another, nor metaphysically continuous with each other.⁶² According to McPherson, to say that property A is metaphysically continuous with property B means, roughly, that B belongs to some category F such that A’s nature can be explained in terms of features that are essential to F (or vice versa). For example, if B is a natural property, then A is metaphysically continuous with B just in case A’s nature can be explained in terms of properties that render B a natural property, such as being spatio-temporally located or being a natural kind.⁶³

The Humean Challenge is meant to be modest in two respects. First, one of its constituent claims, the MODEST HUMEAN thesis, is intended to rule out only brute (i.e., inexplicable) necessary connections between discontinuous properties. Accordingly, McPherson writes: “If two properties are discontinuous but we are nonetheless able to offer a compelling explanation of the necessary connection between them, MODEST HUMEAN makes the reasonable suggestion that having made such a connection intelligible would be enough to defeat the Humean presumption.”⁶⁴ Second, if the nonnaturalist were to posit such brute connections, McPherson acknowledges that this would merely count significantly against the view, but wouldn’t refute it, as nonnaturalism may exhibit other virtues that compensate for this theoretical deficiency.⁶⁵

We would like to offer a pair of replies to the Humean Challenge, neither of which consists in simply attacking the MODEST HUMEAN thesis (of which, for the record, we are deeply suspicious).⁶⁶ Suppose, first, that Minimal nonnaturalism is true: there are nonnatural moral truths, such as the moral fixed points, but there are no nonnatural moral properties. If this view is correct, then the Humean Challenge fails to make contact with it. For the Humean Challenge is aimed at a position according to which there are nonnatural moral properties. Minimal nonnaturalism, however, construes all moral properties as natural ones.⁶⁷

Suppose, second, that Robust nonnaturalism is true: there are both nonnatural truths, such as the moral fixed points, and nonnatural properties. Robust nonnaturalists can elude McPherson’s critique by rejecting BRUTE CONNECTION, claiming that, given everything that McPherson has argued, nonnaturalists needn’t hold that the supervenience of ethical properties on natural ones involves a brute connection between discontinuous properties. For it might be that, when it comes to the supervenience of the moral on the natural, the relevant sort of necessary connection that obtains between natural and moral properties is explicable, since it obtains by conceptual necessity. The moral fixed points offer a model of how to think about such connections, as it might belong to the concept ‘being wrong’ that, given a certain range of natural properties, these properties must have the higher-order moral property being wrong.

Toward the end of his discussion, McPherson anticipates this reply, offering the following reason for rejecting it:

The central point to note in response is that we do not as a matter of language, get to simply stipulate necessary connections between distinct properties. We don’t get to stipulate, for example, that “all Fs are Gs” is analytic, while holding fixed the antecedent meaning of “Fs” and “Gs.” This point can be illustrated by reductio. Suppose that I were permitted to stipulate that I was adding to my idiolect a term “kats” such that “kats” refers to cats, but also that “all kats wear hats” is analytic, while holding fixed the ordinary English meaning of the rest of my language. If it were successful, this stipulation would allow me to truly make the claim that “all kats wear hats.” But this is absurd, since there are evidently cats that do not wear hats. This example suggests a constraint: given an allegedly analytic connection, there needs to be something in the metaphysics that can explain why this connection is necessary … the reductive naturalist has exactly such an explanation.⁶⁸

We agree that we do not get to simply stipulate analytic necessary connections between distinct properties. And we agree—at least for the sake of argument—that there must be “something in the metaphysics” that grounds any such necessary connection.⁶⁹ Were robust nonnaturalists to avail themselves of our position, however, they could satisfy these two constraints.

Our view implies, after all, that if there is a necessary connection between discontinuous properties, such as the natural property recreational slaughter and the nonnatural property being wrong, then this connection holds of conceptual necessity. Since concepts, in our view, are not linguistic (or mental) items, the necessary connection in question would hold independently of any linguistic act, including stipulative ones. Moreover, if there were such a connection, it would be grounded in “the metaphysics,” albeit in the metaphysics of concepts rather than properties. Since, as we see things, conceptual truths hold in virtue of the essences of their constituent concepts, our view provides robust nonnaturalists with precisely the right sorts of resources to respond to the Humean Challenge. (At the very least, this understanding of conceptual truths provides reasons for rejecting the argument that McPherson offers for believing that the relevant sort of necessary connections between the natural and the moral couldn’t be conceptual.) This strikes us as a significant payoff of the approach we’ve developed.

We have argued that our position allows both Minimal and Robust nonnaturalists to respond adequately to the Humean Challenge. Still, the response we’ve offered might strike some as evasive, papering over the fact that our view does posit brute necessary connections between discontinuous entities.⁷⁰ After all, the view we advocate implies that there are necessary connections between natural and nonnatural concepts, which seem to be discontinuous in the sense McPherson has in mind. Is this a problem?

Not so far as we can see. Suppose for the moment that natural and moral concepts are discontinuous in the relevant sense. Now bring to mind concepts such as ‘value,’ ‘promise,’ and ‘God.’ These concepts, with which we are all familiar, are all live candidates for being nonnatural. None, for example, plays any essential explanatory role in the usual sciences. It is, moreover, a matter of considerable controversy whether any of these concepts can be reductively analyzed solely in terms of concepts employed by the usual sciences. (For example, no one, to our knowledge, has offered a satisfactory reductive analysis of illocutionary act concepts such as ‘promise.’) Call a proposition mixed if its elements include what appear to be both natural and nonnatural concepts. Now consider mixed propositions such as:

Values are not sandwiches.

No promise is a quark.

Or:

If God exists, then God is a conscious being.

Each of these propositions appears to be a legitimate candidate for being a conceptual truth. It would, for example, appear to belong to the essence of the concept ‘God’ that, necessarily, if anything falls under it, then it also falls under the concept ‘conscious being.’ But if so, it would be a mistake to reject moral nonnaturalism simply on the grounds that it implies that there are conceptual truths composed of both natural and nonnatural concepts. We ought not to extend the MODEST HUMEAN claim in such a way that it is also a thesis about discontinuous concepts.

7 Objections and replies

Our aim has been to present a version of nonnaturalism that is attractive because it provides the resources for replying, in a pleasingly unified way, to important objections, some of which we canvassed in the last section. Earlier we acknowledged that friends and foes alike would nonetheless be suspicious of our view. We close by considering a series of further objections and offering replies. We believe that these replies throw the attractions of the view we’ve defended into even sharper relief.

• Objection #1: Despite being the subject of considerable controversy, the Open Question Argument has been and continues to be a major motivation for accepting nonnaturalism. This argument, however, hinges on the following assumption:

  • OQA: Given any natural property N, it is always an open question to ask, of any x that is N, whether it is wrong: asking the question “Is an x that is N also wrong?” betrays no conceptual confusion.⁷¹

• However, this assumption is incompatible with what you have claimed about some natural properties, such as recreational slaughter. For you have intimated that if a person sincerely asks whether the recreational slaughter of a fellow person is wrong, then she suffers from a conceptual deficiency. But if so, then you have deprived yourselves of a very important motivation for embracing nonnaturalism. For any decent version of nonnaturalism ought to be at least compatible with the Open Question Argument.

The mesh between this objection and our position is not perfect. For we have not claimed that merely sincerely asking the question stated in OQA betrays conceptual confusion. Indeed, we’ve been at pains to insist that rejecting the moral fixed points needn’t involve conceptual confusion at all. Suppose, however, that we bracket these misgivings for the moment, and instead focus on the point of the Open Question Argument itself. As we understand it, its ultimate aim is to establish that moral concepts (or properties) are not identical with naturalistic ones. An assumption such as OQA is one way in which one might attempt to establish this non-identity thesis. But—and this is the point we wish to emphasize—it is not the only or, in our judgment, the best way to do so.

Here is an alternative and more modest approach that is available to nonnaturalists: we proceed piecemeal by considering some specific proposal for a property identity, such as:

The moral property being morally wrong is identical with the natural property being such as to cause great pain.

We then ask whether an action (epistemically) could cause great pain without being morally wrong. If it could, then this is defeasible evidence that the properties are not identical. This approach does not require one categorically to affirm that, given any natural property N, it is always an open question whether some x that is N is wrong. Thus this approach does not require one to hold that if an action is, for example, a case of slaughtering another human being just for kicks, then it is an open question whether that action is wrong.

It might be worth stressing that the moral fixed points, which we claim are a species of conceptual truth, do not express concept identities. The proposition 〈that it is wrong to engage in recreational slaughter of a fellow person〉 neither states nor implies that the concepts ‘recreational slaughter’ (of fellow persons) and ‘being wrong’ are identical. Nor do these truths express property identities, according to which a natural property such as recreational slaughter is identical with a moral property, such as being wrong. As a dialectical matter, then, one could (in principle) ask whether it is possible for something to be a case of recreational slaughter of a fellow person and not be wrong. But the point of doing so would not be to undermine our proposal that the concept ‘being wrong’ (or the property being wrong) is identical with the concept ‘recreational slaughter’ (or the property recreational slaughter). For we offer no such proposal.

Our position is therefore incompatible with an assumption such as OQA, which figures in some formulations of the Open Question Argument. But not all versions of the Open Question Argument invoke OQA. Versions of the argument that proceed in a piecemeal fashion—which are arguably the most modest and promising versions of the argument—are perfectly compatible with the sort of nonnaturalism that we’ve defended. Thus for the objection under consideration to be persuasive, we would need additional reasons to hold that versions of the Open Question Argument that rest on OQA are more plausible than those that do not. We are not aware of any such reasons.

Objection # 2: Central to your view is the thesis that the moral fixed points are conceptual truths. While you have offered some defense of this view, it is difficult to see how it could be true. Conceptual truths are, after all, supposed to be true in virtue of the essence of their constituent concepts. If that is so, then a proposition such as 〈that it is wrong to engage in recreational slaughter of a fellow person〉 should be true in virtue of the essence of the concept ‘being wrong.’ But insofar as this fixed point is supposed to pertain to beings like us in a world such as ours, it incorporates or presupposes information about human psychology that (among other things) is not an implication of the concept ‘being wrong.’ If so, the fixed points are not good candidates for being conceptual truths.

Our response is to draw attention to a qualification that we made earlier, which is that there are different species of conceptual truth. While there are various ways to distinguish types of conceptual truth, one way to do so is to distinguish immediate conceptual truths, which are true in virtue of the immediate essence of their constituent concepts, from mediate conceptual truths, which are true in virtue of the mediate essence of their constituent concepts. We suggest that the moral fixed points are good candidates for being mediate conceptual truths. We can illustrate this distinction by drawing upon some of Kit Fine’s work.

In his paper “Ontological Dependence,” Fine writes:

We may draw a distinction between the immediate and mediate essence of an item. It is of the immediate nature, or essence, of singleton Socrates [the set containing only Socrates] to contain Socrates and of the immediate nature of Socrates to be a man, but it is only of the mediate nature of singleton Socrates to contain something that is a man. In general, the mediate nature of an object will be subject to chaining: the nature of any object (ineliminably) involved in its nature will also be in its nature. The immediate nature, by contrast, will include only what has a direct bearing on the object, excluding what derives from the nature of the other objects.⁷²

We can expand upon Fine’s thought as follows. It belongs to the immediate essence of singleton Socrates to contain Socrates. Suppose, however, that it belongs to the immediate essence of Socrates to be a human being. And suppose it belongs to the immediate essence of a human being to be a material object. If this is so, it belongs to the mediate essence of singleton Socrates to contain a material object. For we can proceed down the “essentialist chain” to this substantive conclusion.

We earlier claimed that concepts also have essences. If they do, Fine’s distinction can also be applied to them. Suppose—to pick up the chain from the previous example—that it belongs to the immediate essence of the concept ‘human being’ that, if it applies to anything, then it applies to material objects. And suppose it belongs to the immediate essence of the concept ‘material object’ that, if it applies, then it applies only to that which is spatially extended. If this is so, then it belongs to the mediate essence of the concept ‘human being’ that, necessarily, for any x, if x falls under it, then x is spatially extended. Or otherwise put: it is a mediate conceptual truth 〈that human beings are spatially extended〉.

With this in mind, let us turn to the moral fixed points. Consider the proposition 〈that it is wrong to engage in recreational slaughter of a fellow person〉. In our earlier gloss, we said that it belongs to the essence of the concept ‘being wrong’ that, necessarily, if anything falls under the concept ‘recreational slaughter,’ then it also falls under the concept ‘being wrong’ (in worlds like ours). The objection we are considering is that it is very difficult to see how this proposition could be true in virtue of the essence of the concept ‘being wrong,’ since it includes (among other things) broadly empirical information about beings like us.

It should be conceded that the immediate essence of the concept ‘being wrong’—which specifies what it is to be wrong—does not include any information about beings like us. Still, there is a respectable case to be made that the mediate essence of the concept has implications for beings like us. In what follows, we don’t develop this case with the detail it deserves, but simply illustrate how it is that a concept such as ‘being wrong’ might, in virtue of its essence, have implications such that it is plausible to hold that the proposition 〈that it is wrong to engage in recreational slaughter of a fellow person〉 is a mediate conceptual truth.

Begin with this observation: the concept ‘being wrong’ cannot apply to the behavior of plants or non-rational animals. It pertains only to the behavior of rational beings. It is therefore plausible to hold that it belongs to the immediate essence of this concept that it applies only to the behavior of rational agents. The immediate essence of the concept ‘rational agent,’ however, is such that this concept applies to only those beings who meet certain criteria, such as having and being able to execute intentions under concepts that guide their behavior. Call agents who satisfy these criteria intentional agents. Arguably, though, it belongs to the immediate essence of the concept ‘intentional agent’ that this concept must apply to ordinary human beings, for, necessarily, they are intentional agents. Now travel down one more link in the chain: the immediate essence of the concept ‘ordinary human being’ is such that this concept necessarily applies to beings who are mortal, embodied, not subject to daily bouts of complete amnesia, susceptible to physical, emotional and other psychological pleasures and pains, and so forth. In short, it applies of necessity to what, in our presentation of the moral fixed points, we called beings like us in worlds like ours.

This is an illustration of how the relevant chaining might go. In principle, there might be other ways to illustrate it that appeal to the non-moral concepts that constitute the moral fixed points, such as ‘recreational slaughter.’ However that may be, we acknowledge that proceeding down the essentialist chain in the way we have does not itself establish that a moral fixed point such as 〈that it is wrong to engage in recreational slaughter of a fellow person〉 is a mediate conceptual truth. But it would establish how it could be that it belongs to the mediate essence of ‘being wrong’ that, necessarily, this concept pertains to beings like us in worlds such as ours, which addresses the original worry.

Fine’s distinction also helps us to see why certain conceptual truths are more evident than others. Many immediate conceptual truths are highly evident, as grasping their constituent concepts often gives us more or less direct insight into their essence. But many mediate conceptual truths will be much less evident, as grasping them may require us to travel some distance down the essentialist chain. Moral fixed points, such as 〈that it is wrong to engage in recreational torture〉, are plausible candidates for being both mediate conceptual truths and highly evident, since the implications of their constituent moral concepts are there, lying just below the surface.

Objection #3: The moral fixed points, by all appearances, are an “unconnected heap of duties.”⁷³ The examples you provided earlier indicate that they are a hodge-podge of moral propositions with no unifying structure. But moral theory must exemplify a unified and coherent structure; otherwise, the resulting view will lack normative explanatory power and will utterly fail to provide practical guidance for cases that extend beyond those covered by the fixed points.

This objection assumes that the diversity of the moral fixed points will render morality uncodifiable, because the moral fixed points must be understood as fundamental moral truths (i.e., those that are underivable from broader moral principles).

We reject this assumption. Our view contains no commitment to seeing the moral fixed points as fundamental moral truths. For all that we have said, some of the fixed points that we have identified may be instances of and derivable from yet more general and fundamental moral fixed points, of the sort, for instance, expressed in Ross’s principles of prima facie duty or in the possibly indefeasible proposition 〈that it is wrong to visit harm on other rational agents merely for the sake of the pleasure it provides〉. Indeed, our position is compatible with the view that all of the moral fixed points may be unified by virtue of a single supreme moral principle, or perhaps a coherent and ordered set of more general principles.⁷⁴

As to the extent to which morality is codifiable, we believe that at present this is an open question, thinking it best not to make ex ante assumptions about the degree of systematicity exemplified by the set of first-order moral truths. We are not completely ecumenical here—our endorsement of the Reversal Argument does, after all, entail a rejection of ethical particularism. But that is not much of a restriction in this context. Our position is otherwise compatible with any view of the structure of moral theory.

Still, one might object that if there were some deeper unifying principle from which the moral fixed points could be derived, then their truth would be explained not by the essence of their constituent concepts, as we have claimed, but rather by this supreme moral principle. In this case, the wrongness of, say, recreational slaughter or gratuitous humiliation would be explained not by the fixed points, but instead by the supreme moral principle from which the fixed points are derived.

This objection, however, does not genuinely engage our view. To see this, suppose that there is some deeper unifying principle from which the moral fixed points could be derived. And suppose, for argument’s sake, that this unifying principle is itself a conceptual truth. These commitments are not in tension with our position. Recall that, as we understand it, a conceptual truth is a proposition that is true in virtue of the essences of its constituent concepts (and not some correlative worldly fact). This formulation does not say or imply that a conceptual truth is a proposition that is true solely in virtue of the essences of its constituent concepts. Our view is thus compatible with a position according to which we can derive the moral fixed points from other normative principles. Indeed, our discussion of immediate and mediate conceptual truths turned on the recognition of something close to this very point.

In sum: whether the moral fixed points can be unified under the banner of one of the traditional moral theories, or one not yet discovered, is a matter that can be determined only after a great deal of substantive normative theorizing. All that we insist upon is that such theorizing respect the role that the moral fixed points play—that of serving as the conceptual boundaries of a minimally eccentric moral system. Whether such theorizing proceeds by subsuming the fixed points under one supreme moral rule or a battery of more general moral principles, or by regarding them as fundamental truths, is a matter on which we prefer to remain neutral.

Objection #4: Nonnaturalists have long assumed that moral facts and properties are causally impotent. But if we cannot explain our moral beliefs by citing our causal contact with moral facts and properties, then there is no plausible explanation of how our moral faculties could be reliable. So your view, like other versions of nonnaturalism, lacks a credible moral epistemology.

One way to counter this objection would be to show that nonnaturalists can accommodate the causal efficacy of moral facts and properties. While we have some sympathy for this view, suppose for argument’s sake that moral facts and properties possess no causal powers.⁷⁵ If that is so, then our moral beliefs are not caused by moral facts. And if that is so, then it can seem mysterious how we might reliably discern them.

But we can dispel the apparent mystery, at least when it comes to those beliefs that have the moral fixed points as their content. That is because the moral fixed points are a species of conceptual truth, and it is highly implausible to insist on causal conditions for obtaining knowledge of such truths. Consider the fact that actuality entails possibility. This fact is causally impotent. Still, we know that actuality entails possibility. We know it because we have an adequate grasp of the concepts ‘actuality’ and ‘possibility’ and have formed our belief in the proposition 〈that actuality entails possibility〉 on the basis of that adequate understanding. We hold this belief as a result of exemplifying a degree of conceptual mastery that establishes a non-accidental, reliable link with the truth. That yields a level of warrant sufficient for knowledge.

We can tell the very same story when it comes to the moral fixed points. Suppose that the fact that it is wrong to engage in recreational slaughter of a fellow person is causally impotent. A fortiori, this fact is not causing us to believe that engaging in recreational slaughter of this sort is wrong. But this is no bar to our knowing that such killing is immoral. For the proposition 〈that it is wrong to engage in recreational slaughter of a fellow person〉 is a conceptual truth. So long as, in grasping this proposition, we have an adequate understanding of the concepts ‘recreational slaughter’ and ‘being wrong’ (and their relations), then we will have arrived at our belief in a reliable manner.

We are well aware that some philosophers reject the existence of conceptual truths, and as a result will deny that there can be conceptual knowledge of the sort that we are defending. Still, if we place such skepticism to the side, nonnaturalists can take over whatever account of conceptual knowledge is best, and deploy that account in the service of explaining how knowledge of the moral fixed points is possible. So, while we have not furnished anything remotely like a complete nonnaturalist moral epistemology—a tale must be told, for instance, of how we can come to know moral propositions other than the fixed points—still, we can see how some (and perhaps much) moral knowledge is within reach, even if moral facts and properties are causally inert.

Objection #5: Your characterization of those who reject the moral fixed points is uncharitable in the extreme. Error theorists and immoralists can be subtle, even brilliant, thinkers; to accuse them of conceptual deficiency in moral matters is implausible on its face.

In handling this charge, it is helpful to consider error theorists and immoralists separately. Let us begin with error theorists.

When we say that error theorists are mistaken in their views because they fail to affirm propositions that we regard as a species of conceptual truth, we are not thereby committed to attributing to them simple-minded positions or flat-footed philosophical mistakes. Rather, our claim is that error theorists are failing to recognize a set of conceptual truths as a result of having been convinced by sophisticated, albeit unsound, philosophical arguments. Assessing the force of such arguments is anything but straightforward (which is why we nonnaturalists ought to be far less than fully confident of our own views). Those who come down on what we regard as the wrong side of those arguments needn’t be making any silly or obvious mistakes.

That said, we do find the error theorist’s methodology problematic. The mistake that error theorists make, in our view, is not simply a failure to appreciate or acknowledge certain conceptual truths. It also consists in rejecting highly evident first-order moral propositions such as:

〈that it is wrong to engage in recreational slaughter of a fellow person〉

on the basis of either highly controversial metaethical claims such as:

There are no categorical reasons,

or speculative empirical claims such as:

Our moral judgments are thoroughly contaminated by non-truth-relevant influences, such as various evolutionary pressures,

which figure in prominent objections to realism.⁷⁶

There is, admittedly, some irony in this diagnosis of the error theorist’s situation, for philosophers frequently accuse nonnaturalists of loading first-order moral judgments with unnecessary metaphysical baggage, such as the thesis that some moral propositions are nonnatural moral truths. Nonnaturalists, we think, should own up to the fact that theirs is a view that has distinctive ontological commitments that require justification. But they should deny that their ontological commitments are the product of employing a suspect philosophical methodology. It is their rivals, after all, who are typically willing to jettison extremely plausible first-order claims such as the moral fixed points in the service of advancing highly contentious metaethical positions, such as those mentioned above.

When it comes to immoralists—those who endorse a set of putatively moral norms that imply the rejection of the moral fixed points—we will have to offer a different diagnosis. Immoralists advocate for normative systems whose constituent principles are deeply heterodox. Think of someone like Nietzsche in certain of his moments, intent on elevating the importance of power, celebrating humiliation and the subjugation of the humble and the weak, denying that compassion or kindliness are virtues. Almost everyone shares our distaste for the immoralist’s favored norms. But our critics will object that we are making things too easy on ourselves and too quickly ruling out immoralism when we insist that its proponents are committed to views that are conceptual falsehoods.

We think that this concern can be met. To see how, imagine that someone were to maintain that recreational slaughter is good or admirable. If we confirm that the person is sincere, then we are forced to one of two diagnoses.

In the first place, the person might be using these epithets in a nonmoral sense. In commending an action that we regard as morally repugnant, the immoralist may be utilizing evaluative language to indicate that such an action is, say, likely to promote his selfish ends or those of a group intent on harming others. In such cases, moralists and immoralists are using the same words to express different concepts, and they will not, in fact, be engaged in a moral disagreement at all. Still, they may well be engaged in a normative disagreement, since this kind of disagreement can usually be translated into one about which course of action we have excellent or most reason to pursue. In this scenario, we might be able to explain the possibility of serious normative (albeit nonmoral) disagreement, without attributing any sort of conceptual error to the immoralist.⁷⁷

Second, if immoralists really are intending to construe these favoring predicates as moral ones, and to apply them to actions such as recreational slaughter and rape, then there will indeed be plenty of room for disagreement here: we take the moral fixed points to be true, and immoralists don’t. Admittedly, in this second scenario, we must charge immoralists with having made a conceptual error. But such a charge does not in this case strike us as deeply uncharitable.

Almost everyone, after all, rejects immoralism. It isn’t clear why those (like us) who do so on the basis of views about the nature and identity of conceptual moral truths are ipso facto any less charitable than those who reject immoralism on the basis of what they regard as substantive, non-conceptual moral truths. We attribute to immoralists both conceptual and moral deficiencies; other moralists will attribute “merely” a variety of epistemic and moral deficiencies. The explanation we offer of the shortcomings of immoralism is compatible with assigning to immoralists a very high degree of general intelligence, knowledge of nonmoral facts, conative coherence, and instrumental rationality.

It might be that the underlying worry here is that if something is a conceptual truth, then it should garner consensus among those possessed of a high degree of intelligence and rationality. On this line, the fact that brilliant immoralists dissent from conventional moral wisdom is evidence enough that the moral fixed points are not a species of conceptual truth.

There is much to say here, but rather than extend an already lengthy effort, we should remind ourselves of a point made earlier. Not all conceptual truths are obvious (though we do think that the moral fixed points are evident to a very high degree). If someone disagrees with a claim that we regard as a conceptual truth, then we owe an account of the conceptual deficiency. Such explanations will differ on a case by case basis, but the familiar types, cited above in our discussion of doxastic blindspots, should suffice to do the needed work here.

We take our leave with a challenge. There are many ways to set the conceptual parameters of a minimally eccentric moral system. We prefer to do this in a way that imposes contentful conceptual constraints. Others do not; the history of moral philosophy contains many alternatives to our position, such as those that delimit a moral system by reference to the norms that ideal observers would endorse, or those that are justifiable to all, or those whose (near) universal endorsement would maximize social utility, and so forth. If immoralists are really seeking to offer an account of morality, as opposed to supplanting it with a different kind of normative system, then the burden falls squarely on them to defend a conceptual framework for identifying the correct moral principles, a framework that in fact yields the norms that immoralists most favor. We are not sanguine about the prospects of accomplishing this task.